J. Sens. Sens. Syst., 6, 9–18, 2017www.j-sens-sens-syst.net/6/9/2017/doi:10.5194/jsss-6-9-2017© Author(s) 2017. CC Attribution 3.0 License.

Ultrasound tomography for inline monitoringof plastic melts

Norbert Halmen, Christoph Kugler, Thomas Hochrein, Peter Heidemeyer, and Martin BastianSKZ – German Plastics Center, Würzburg, 97076, Germany

Correspondence to: Norbert Halmen (n.halmen@skz.de)

Received: 11 August 2016 – Revised: 30 November 2016 – Accepted: December 2016 – Published: 10 January 2017

Abstract. The inline determination of process and product parameters is of great help for the evaluation andoptimization of new procedures. Therefore, an ultrasound process tomography system has been developed, whichenables the imaging of the local filler distribution in plastic melts.

The objects investigated were extruded rods made of polypropylene (PP) with radial filler gradients. Duringextrusion, sound velocity and attenuation of the plastic melt were determined and processed via a modifiedreconstruction algorithm according to Radon transform into 2-D sectional images. Despite the challenges ofhigher attenuation and impedance mismatch of 60 mm filled PP melt compared to water, the resulting images areof good quality. An important factor for the image quality after tomographic reconstruction is the opening angleof the used ultrasound transducers. Furthermore, a simulation environment was developed in Matlab, whichserves as a testing platform for the measurement system.

1 Introduction

1.1 Ultrasound for process monitoring

Ultrasound (US) waves are mechanical, acoustical – bothshear as well as compressive – waves in the frequency rangefrom 20 kHz up to several GHz, depending on the elastic anddissipative properties of the medium. There are two basicproperties which are used for measurements. First, the speedof sound c is given by Eq. (1),

c =1x

1t, (1)

where 1x is the traveled distance and 1t the propagationtime. Second, the attenuation coefficient α can be defined bythe intensity ratio of original (I0) and attenuated (I ) signal:

α =1

21xlnI0

I, (2)

where the pulse intensity I is

I =

t2∫t1

A2 (t) dt. (3)

Here, A is the signal amplitude and t1 and t2 are the start andend time of an US pulse. In general, when examining ma-terials with US, advantages can be taken of the frequency-dependent attenuation due to viscous losses as well as scat-tering and thermoelastic effects while sound waves are prop-agating through a medium. Since US has been establishedmainly in the field of non-destructive testing and for diag-nostic purposes in the medical sector, it has also arrived as aprocess and quality monitoring tool in the plastics sector. Inliquids and plastic melts, US waves propagate mainly in theform of longitudinal waves. The material properties dependon the complex longitudinal modulus M∗, given by Eq. (4):

M∗ =M ′+ iM ′′ =%c2

(1− iαcω

)2, (4)

where % is the density, c is the speed of sound, α is the at-tenuation coefficient and ω is the angular frequency, which isdefined by

ω = 2πf. (5)

Since the attenuation in plastic melt is small in most cases,the real part of the longitudinal modulus can be simplified by

M ′ = %c2. (6)

Published by Copernicus Publications on behalf of the AMA Association for Sensor Technology.

10 N. Halmen et al.: Ultrasound tomography for inline monitoring of plastic melts

Then, the speed of sound c can be easily calculated, which iscomposed of the real part of the bulk modulus K ′ and shearmodulus G’:

c =

√M ′

%=

√K ′+ 4

3G′

%. (7)

Because of the very small shear modulus, Eq. (7) can be sim-plified in the following way:

c =

√K ′

%. (8)

Equation (8) shows the pressure dependence of the speed ofsound, since K’ and % are pressure dependent. The genera-tion of US waves is mostly based on the piezoelectric effect(Deutsch et al., 1997). Near-real-time measurements (Dinger,2007) qualifies US as a non-destructive testing procedure forinline measurements in plastic melts.

In transmission arrangement (i.e., transmitter and receiverare arranged opposite to each other) the pulse is attenuatedwhen passing through the melt and the propagation time de-pends on the polymer. This allows conclusions about the vis-cous and elastic properties of the plastic melt. Especially sig-nals with frequencies between 100 kHz and 10 MHz can alsopropagate effectively in highly filled compounds (Alig et al.,2000) – but only as long as the size of the filler particles (scat-terers) is much smaller than the wavelength. Previous studieshave shown that the volume fraction of different filler mate-rials in different base polymers can be determined by vari-ations in sound velocity (Alig et al., 2000, 2005). However,this depends on pressure and temperature. In the same man-ner, the distribution of particles also has an influence on theattenuation of the US signal. Therefore, the filler size distri-bution and agglomerate formation can be determined as well(Alig et al., 2006; Schober et al., 2014; Wöckel et al., 2016).

Since piezoelectric materials are temperature sensitivethey need to be protected from high temperatures. For thisreason testing in hot media needs specially designed UStest probes protected by appropriate measures (Putz et al.,2014). High temperature environments require delay lineswhich allow a thermal decoupling between the temperature-sensitive US transducers with their adhesive joints and thehot melt. Such delay lines are made of materials which pos-sess good sound transmission but low heat conductivity prop-erties, such as glass, ceramics, special steel alloys, or high-performance plastics (Deutsch et al., 1997). When soundwaves propagate through interfaces between different materi-als the reflection coefficient r , defined as the ratio of reflectedto incident amplitude of a sound wave, has to be considered.It is given by Eq. (9):

r =Z1−Z2

Z1+Z2, (9)

where Z1, Z2 are the material impedances of adjacent mate-rials. The impedance of a material is calculated by

Z = %c. (10)

Equation (9) shows that impedance matching is very impor-tant to minimize reflection losses. This implies the impor-tance of the right material choice. When working with poly-mer melts, materials like polyimide (PI) or polyetheretherke-tone (PEEK) can be used as coupling materials at tempera-tures of up to 250 ◦C.

1.2 Polymeric gradient materials

When extruding semi-finished plastic products, the sub-sequent component properties are intentionally modifiedthrough incorporating additives, for example, to fulfill spe-cial requirements on mechanical properties and resistanceagainst light or chemicals. In this case, the functional ad-ditives are typically homogeneously distributed across theentire extruded profile. An intentional distribution of aggre-gates in the component can either be accomplished throughco-extrusion in the form of sharply defined layer systemswith concentration jumps on the boundary layers (Zhu et al.,2006) or through discontinuous procedures (Wu et al., 2004;Sun et al., 2008; Krumova et al., 2001). A continuous pro-cedure for the extrusion of rods made of polypropylene (PP)with radial gradients of fillers is currently being developed(Hirt et al., 2015). As an optimal adjustment of the gradient(no local displacements, no agglomeration of the additives)is decisive for the component properties of the final semi-finished product and allows unambiguous conclusions aboutthe quality of the production process, the continuous moni-toring of the extrusion process is of advantage. Typically, theexamination is offline and takes place in the lab after pro-duction either visually by sample or microscope or with theuse of X-ray computed tomography (CT). However, offlinemethods entail a long dead time in the area of hours betweensample-taking (including cooling of the rod) and the presenceof a measurement result, which makes the process develop-ment and optimization very time-consuming. Furthermore,very small differences in density (e.g., as in the case of poly-mer blends) cannot be resolved by X-ray CT.

1.3 Ultrasound process tomography (USPT)

For an inline determination and real-time process control anUSPT system has been developed for an inline imaging ofthe melt during the extrusion of rods. In contrast to X-rayCT, US requires no safeguards and can be integrated into theprocess directly.

Comparable approaches already exist in the medical sec-tor for mammographies (Schwarzenberg, 2009), in the plas-tic sector, for example, to dissolve the temperature distribu-tion in a melt channel (Praher et al., 2014; Hopmann et al.,2016), or for non-destructive component testing (Haach and

J. Sens. Sens. Syst., 6, 9–18, 2017 www.j-sens-sens-syst.net/6/9/2017/

N. Halmen et al.: Ultrasound tomography for inline monitoring of plastic melts 11

Ramirez, 2016). An US tomography-based application forspatially resolved determination of the filler distribution inplastic melts has not been reported so far.

2 Experiment

2.1 Measurement system and materials

The USPT system was designed to the geometry of rods andconsists of a sensor ring with an inner diameter of 60 mm, al-lowing an easy adoption on the melt channel right before theexit of the shaping nozzle (see Fig. 1). The used basic poly-mer was Sabic 505P, a PP with a density of 905 kg m−3 and amelt flow rate of 2 g/10 min (230 ◦C, 2.16 kg). For the fabri-cation of a black to white color gradient, two different fillerswere used: on the one hand 0.18 % carbon black (SR 301Carbon Black of Sid Richardson Carbon & Energy Co., par-ticle size 30–50 nm, density 1700–1900 kg m−3 at 20 ◦C) andon the other hand 2.2 % titanium dioxide (TiO2 2220 of Kro-nos Titan GmbH, mainly rutile, particle size 150–250 nm,density 4000 kg m−3). The longitudinal speed of sound insolid graphite is 1470 m s−1 at 20 ◦C (Samsonov, 1968),and for titanium dioxide the value varies from 9070 m s−1

(277 ◦C) to 9240 m s−1(27 ◦C) (Munro, 2002). These valuesare significantly greater than the typical speed of sound forpolymer melts of about 1000 m s−1. The melt temperatureduring processing ranged from 200 to 230 ◦C, and the pres-sure was around 1 bar. Moreover, measurements were con-ducted on distilled water.

The speed of sound in media depends on temperature. Ifthe material parameters are known, the speed of sound canthen be derived from (Hochrein and Alig, 2011)

ϑmelt = Cϑ(c− v0−Cp (p− 1bar)

)+ ϑ0, (11)

with speed of sound c, melt temperature ϑmelt, medium spe-cific reference temperature ϑ0, reference speed of sound v0,temperature compensation coefficient Cϑ and pressure com-pensation coefficient Cp. As the pressure in our setup is closeto 1 bar, Eq. (11) can be simplified to

ϑmelt = Cϑ (c− v0)+ ϑ0. (12)

Using Eq. (12) and literature values for Cϑ , ϑ0 and v0 for PP(Hochrein and Alig, 2011) for the temperature range from180 to 260 ◦C, the speed of sound can be calculated as fol-lows:

c =ϑmelt− ϑ0

Cϑ+ v0 =

ϑmelt− 473.15K

0.50 Ksm

+ 954ms. (13)

Due to the material for the delay lines, the impedance mis-match at the interfaces to water and PP has to be taken intoaccount. At 25 ◦C the speed of sound is 2900 m s−1 and thetemperature gradient is −0.6 (m s−1) K−1 and decreases lin-early with increasing temperature. This correlation between

Figure 1. View on the delay lines of the US transducers inside ofthe USPT adapter (top) and USPT adapter (wrapped in yellow poly-imide tape) mounted between gradient tool and shaping nozzle ofthe used extrusion line (bottom).

the speed of sound and the temperature for the delay line ma-terial is valid from 25 to 200 ◦C. The density of the materialat 25 ◦C is 1300 kg m−3. Since the high temperature materialdoes not change its density in a wide range even at highertemperatures, the given density is used for the calculation ofthe impedance and reflection coefficients at 22 ◦C for waterand 220 ◦C for PP, listed in Table 1. It can be seen that forwater the reflection losses are 17 % lower due to the betterimpedance matching, resulting in higher signal intensities.

The attenuation coefficient of sound in water can becalculated using the k-Wave toolbox (Treeby and Cox,2010) in Matlab. For a temperature of 22 ◦C and a fre-quency of 2.12 MHz the attenuation of water is calculated toαH2O = 0.0093 dB cm−1. According to Schober et al. (2014)the attenuation coefficient for PP melt (αPP ) at the same fre-quency can be estimated to 0.3–0.6 dB cm−1 or even slightlyhigher because of higher temperature and lower pressurecompared to the setup of Schober et al. (2014). Compared

www.j-sens-sens-syst.net/6/9/2017/ J. Sens. Sens. Syst., 6, 9–18, 2017

12 N. Halmen et al.: Ultrasound tomography for inline monitoring of plastic melts

Table 1. Estimation of the reflection coefficients r for the delay line–water interface and the delay line–PP melt interface at differenttemperatures (%delayline = 1300 kg m−3).

Medium T % v Z Vdelayline Zdelayline r

(◦C) (kg m−3) (m s−1) (106 Ns m−3) (m s−1) (106 Ns m−3)

Water 22 998 1489 1.49 2902 3.77 0.433PP 220 905 994 0.900 2783 3.62 0.602

to water, the attenuation coefficient of PP melt is about 30 to60 times higher. The attenuation in the delay line material isspecified with ca. 4 dB cm−1 for the given frequency.

Due to the higher impedance mismatch and higher signalattenuation of the plastic melt, high transmission powers ofthe US transducers are required to pass through a diameter of60 mm. For a good tomography imaging of the melt channel,each location point has to be scanned from as many direc-tions as possible (Kak and Slaney, 2001). This can be ac-complished either with a large number of transducers or witha wide radiation angle per transducer. With the given geom-etry, a large number of transducers would be synonymouswith smaller apertures and therefore also larger opening an-gles. However, the maximum sound power per transducer de-creases with its aperture due to the smaller contact surface tothe melt.

The USPT was equipped with 40 equiangular trans-ducers with plain apertures of 2 mm× 14 mm× 14 mm(width× length× depth), of which the delay lines are indirect contact with the melt (see rectangular insets in theadapter in Fig. 1). The narrow width orthogonally to the flowdirection is to accomplish a greater opening angle in the sec-tional image plane of the tomography. The greater expansionalong the flow direction of the melt causes only a very smallaperture vertically to the sectional image plane but enlargesthe contact surface of the transducers and thus the obtainablesound power per transducer. The delay lines were glued ontothe piezo crystals using a two-component Epoxy adhesive.Figure 1 shows the sensor system from the inside as well asthe adapter mounted to the gradient tool.

In contrast to typical tomographic methods, the projectionangle is not varied by a mechanical rotation of transmitterand detector but electronically. The transducers in the dif-ferent angular positions are successively set to transmittingmode while the other 39 transducers are in receiving mode. Ifthe attenuation of the investigated medium (e.g., plastic melt)is high, up to four transducers can be bundled and used as atransmitter array. Through phase-delayed controlling of sev-eral transducers a focusing effect can be achieved and even aswivel of the sound field is possible.

The US transducers were driven with a rectangularwindowed sinusoidal signal with a center frequency off = 2.12 MHz and a transmit voltage that can be changedstepwise from 12Vpp to 102Vpp. Moreover, the number ofburst cycles is also adjustable. Investigating water, the ex-

Figure 2. System electronics of the USPT system (Inoson PCM12343).

periments were done with 1 or 3 bursts, while for PP melt5 or even 10 bursts were used to achieve higher signal in-tensity. The US energy is emitted in the shape of a conealong the transducer axis. The maximum intensity is alongthe 0◦ axis and decreases with increasing angle. Simulationsdone by the manufacturer resulted in about 6 dB lower signalintensity – so approximately only a quarter – at an angle of30.3◦. This value was defined as opening angle. The adapterand the transducers are designed for the production of gradi-ent materials working at temperatures of up to 250 ◦C and apressure of up to 20 bar. Aside from the control of the trans-ducers, the electronics of the system (Inoson PCM 12343, seeFig. 2) allows the collection of measurement signals with anADC sampling rate of 12.5–100 MHz and 256–16 384 sam-ple points per acquired amplitude scan (A-scan). The gainlevel for every individual receiver is settable in 255 steps upto 80 dB, and two +12 dB software boosters can be used ad-ditionally. Fast functional testing and optimization of indi-vidual signal parameters were performed using the manufac-turer software based on LabView. For tomographic measure-ments, beginning with signal acquisition through the analysisto the presentation of the results, a graphical user interfacewas developed using Matlab. The time to complete an entiremeasurement including all transducers with a sampling rateof 25 MHz and 8192 sample points is approximately 8 s. Ithas been found that these parameters were the best compro-mise of resolution and speed.

J. Sens. Sens. Syst., 6, 9–18, 2017 www.j-sens-sens-syst.net/6/9/2017/

N. Halmen et al.: Ultrasound tomography for inline monitoring of plastic melts 13

2.2 Calibration

All delay lines for all US transducers in the system wereproduced manually, so geometric differences may occur. Thelength of the delay line in combination with its temperaturedependence directly affects the measured sound velocity. Inaddition, the US transducers can exhibit different transmit-ting/receiving properties depending on the adherence of thepiezo crystals to the delay line. Taking these factors into con-sideration, the system has to be calibrated accordingly. Dis-tilled water was used as calibration medium since the speedof sound and the attenuation are well-known for defined tem-peratures.

For an absolute determination of the speed of sound withinthe USPT, the propagation time offset caused by the delayline of each transmitter and receiver pair (TRP) is necessary.With the measured distances between each TRP this time off-set can be calculated for a given temperature. When using thesystem at higher temperatures it has to be adapted.

The calibration of the sound attenuation for this system is abit tricky, since the transducers cannot directly be comparedto one another by an opposing measurement setup. There-fore, a custom-made omnidirectional US transducer made byInoson GmbH (see Fig. 3) was used in the center of the USPTsystem. The transducer has a diameter of 10.3 mm and wasconnected to a channel of the PCM. In transmission mode thetransducer was driven with 2.12 MHz and a transmit voltageof 47Vpp, while in receiving mode a gain of 67 % was set.Considering the intensities of the transmission pulse I0 andecho (or reference) pulse I along with their distances x0 andx (= 3 x0 in case of echo pulse) the US attenuation can becalculated using Eq. (2). Each individual US transducer ofthe USPT system is assigned with a corrective factor, whichmakes them comparable to one another. Using one trans-ducer as a reference these corrective factors can be appliedon the gain level settings. The material and impedance ofthe omnidirectional transducer are unknown, but since thesecorrective factors are only relative, those properties are notnecessary. Once the gain levels have been set, also the an-gle dependence and pulse profile of every transducer can bedetermined.

2.3 Reconstruction algorithm

When using fillers, scattering effects at the particles has tobe considered. Here the wavelength λ is the decisive factor,which is defined as a quotient of speed of sound c and fre-quency f :

λ=c

f. (14)

For an US frequency of f = 2.12 MHz and speed of sound ofc (220 ◦C)= 994 m s−1, the wavelength is about λ= 470 µm.Since the particle sizes given in Sect. 2.1 are smaller than0.2 λ (94 µm) scattering is described by Rayleigh scatter-ing. Therefore a ray-like propagation was assumed, where

Figure 3. Custom-made omnidirectional US transducer made byInoson GmbH.

the wave on the direct way from the transmitter to receiverreaches the detector first. Scattered waves and reverberationsare delayed and have smaller intensities. Furthermore, refrac-tion and diffraction in polymer melts can be neglected in thefirst approximation. The used wave detection setup of a trans-mitter and opposing receivers is comparable to a fan beamwith a certain opening angle. This was assumed for the cal-culation of the projections from different viewing angles.

To receive a sectional image from the individual pro-jections, a classic back projection algorithm for a non-diffracting case according to Radon (Kak and Slaney, 2001)was adopted.

Only direct paths between transducers are considered, andthe line integral for each TRP was calculated by

Lr,t =

∫line

f (x,y) ds, (15)

where f (x,y) is either the speed of sound or the attenuationat a certain pixel. In order to use the time-optimized func-tion provided by Matlab for back projection reconstruction(inverse Radon transform) the fan beam geometry has to beconverted into a parallel beam geometry (rebinning) since thefunction requires parallel beam projection data. Thereby, animage of all projections over one viewing angle is received.This is known as Radon transform or sinogram and definedby

Pθ (z)=

∞∫−∞

∞∫−∞

f (x,y) δ(x cosθ + y sinθ − z) dxdy, (16)

where θ is the projection angle and z the dimension of theprojection. Because of the Fourier slice theorem, implyingthat the one-dimensional Fourier transform of a parallel pro-jection is equal to a slice of the two-dimensional Fourier

www.j-sens-sens-syst.net/6/9/2017/ J. Sens. Sens. Syst., 6, 9–18, 2017

14 N. Halmen et al.: Ultrasound tomography for inline monitoring of plastic melts

transform of the original object, we can estimate the ob-ject function by a two-dimensional inverse Fourier transform(Kak and Slaney, 2001).

With the inverse Fourier transform,

f (x,y)=

∞∫−∞

∞∫−∞

F (u,v)ei2π (ux+vy)dudv, (17)

where F (u,v) is the Fourier transform of f (x,y). Furtherdetails can be found in Kak and Slaney (2001). Equation (17)can be transformed to

f (x,y)=

π∫0

Qθ (w) (x cosθ + y sinθ ))dθ, (18)

with the projectionQθ (w). Equation (18) shows that the pro-jections over different angles are summed up to the objectfunction, which is called back projection.

2.4 Simulation

Beside the measurement control and analyzation software,an additional simulation environment was developed in Mat-lab. Here, images, in which the different sound velocities andattenuation factors are depicted by different pixel values, areconverted into matrices. The pixel values are integrated alongthe respective pixel line between a TRP. For the simulationthe opening angle can be varied, which determines the num-ber of opposing receivers used for reconstruction. By this,defective receivers can be simulated. In order to generate vir-tual data for the empty areas between the individual (and fordefective) transducers the values were linearly interpolated.The reconstruction was done by creating 2-D sectional im-ages with lines between TRPs of the integrated pixel valueand subsequently summing them up. Afterwards a weight-ing function was applied, which took into account how oftenlines passed through a certain pixel.

The simulation enables the verification of real results, andthe simulation algorithm can be improved on the basis of realmeasurement data. Thus, the suitability of the used USPTsystem can be determined beforehand for specific conditions(defective transducers, opening angle) and material systems.Moreover, the use of different corrective features, such as in-terpolation or filtering of the sinogram in the Fourier space(filtered back projections), can be examined.

2.5 Measuring procedure

The measuring procedure with real media is performed anal-ogous to the simulation. Therefore, the same algorithm withsome extensions concerning signal acquisition and data pre-processing is used. The measurement data are acquired withthe PCM. The recorded A-scan signals (see Fig. 4) are band-pass filtered (finite impulse response) at the center frequency

Figure 4. Bandpass-filtered A-scan of an US signal (53VPP, 3 burstcycles, gain level 49 %) of two opposing transducers in the USPTadapter filled with distilled water: the transmission pulse and thefirst back wall echo can be clearly distinguished and are marked inred.

f with a relative bandwidth of 14 % to suppress interferingnoise signals. A main criterion was also to avoid overdrivingthe transducers and clipping by setting to high gain levels.Consequently, the propagation time is calculated for everysignal impulse based on threshold detection algorithm, whichindicates when the US pulse has reached the receiver. Thethreshold must be adjusted to the signal-to-noise ratio (SNR)depending on the medium. Since the melt in the adapter isvery homogenous when the extrusion process has becomestable, an improvement of the SNR through repeated mea-surements can be observed.

While only one transducer emits, every other transducerreceives a signal. This allows the use of a high number ofreceivers for the reconstruction, resulting in a bigger open-ing angle. Initially, it has to be checked which receivers canbe used for transmission. This can be done by comparing themaximum normalized signal amplitudes for every receiverwhile one transmitter is working. Figure 5 shows the compar-ison of the maximum normalized amplitudes of the receiverswhen transmitter 1 emitted for water and PP at different set-tings. Receiver 21 opposes transmitter 1. Based on severalexperiments, an opening angle of 60◦ was chosen for the fur-ther work.

The propagation time together with the known distance ofthe TRP allows the calculation of the sound velocity alongthat particular path. The intensity of the pulse train is usedto determine the attenuation. However, a measured referencepulse train is required to do so. This can either be a backwall echo (see Fig. 4) or a transmission pulse from a ref-erence measurement. Optionally, by interpolation betweenmeasured values of adjacent receivers, virtual values can begenerated for areas without US transducers or for defectivetransducers. Subsequently, the sinogram is created and a sec-

J. Sens. Sens. Syst., 6, 9–18, 2017 www.j-sens-sens-syst.net/6/9/2017/

N. Halmen et al.: Ultrasound tomography for inline monitoring of plastic melts 15

Figure 5. Comparison of maximum normalized signal amplitudefor every receiver when transducer 1 emitted for water (53VPP, gainlevel all receivers 49 %, 3 burst cycles) and PP melt (102VPP, gainlevel all receivers 69 %, 10 burst cycles, +12 dB booster). The datapresented here are raw data without corrective factors.

tional image can be reconstructed depicting the distributionof the sound velocity or the attenuation within the USPT sys-tem.

3 Results

3.1 Simulation of agglomerates and gradients

The effect of different opening angles of the transducers aswell as the use of an interpolation on the reconstruction is ex-emplarily demonstrated on the basis of simulation images ofagglomerates in Fig. 6 and a gradient of filler concentrationoffset from the symmetry axis in Fig. 7. The “original” pic-tures were created as grayscale images with maximum pixelvalues (255) for the agglomerates and the upper value of thegradients (the lowest was 0). The algorithm itself uses an-other color mapping. It can be perceived that the tradeoffmade during the setup of the measurement system betweennumber and aperture of the transducers leads to tomographicartifacts, such as ring artifacts, shadows, or blurring. Theseeffects are removed through appropriate filtering and adjust-ments, improving the quality of the image.

The imaging resolution is limited by the width of the UStransducers to 2 mm. Due to the limited opening angles of thetransducers the effective area for a meaningful reconstructionis restricted. The larger the opening angle the larger the areawith good reconstruction results. Also the use of interpola-tion seemed helpful. Both effects are also demonstrated inFigs. 6 and 7.

Figure 6. Simulation of a reconstruction of spatially distributed ag-glomerates for different opening angle and optional interpolation(int).

Figure 7. Simulation of a reconstruction of a de-centered gradientfor different opening angle and optional interpolation (int).

3.2 Measurements on filled PP melts

The time for a single measurement with the typical settingsmentioned in Sect. 2.1 is 0.2 s per transducer. For a full to-mographic measurement using all 40 transducers, the wholemeasurement time sums up to 8 s. Within the extrusion pro-

www.j-sens-sens-syst.net/6/9/2017/ J. Sens. Sens. Syst., 6, 9–18, 2017

16 N. Halmen et al.: Ultrasound tomography for inline monitoring of plastic melts

Figure 8. Bandpass-filtered A-scan of an US signal (102VPP,10 burst cycles, gain level 69 %, +12 dB booster) of two oppos-ing transducers in the USPT adapter of a PP melt filled with TiO2.The transmission pulse is marked in red.

cess, the polymer melt continuously flows during the mea-surement with a melt flow velocity of vmelt = 1.25 mm s−1.In this case the covered distance of the melt is about 1 cm,thus 0.25 mm per active transmitter. Because of the move-ment during the measurement a Doppler effect has to beconsidered. The frequency shift 1f is given by (Rosen andGothard, 2010)

1f =vmelt

cPPf, (19)

with the center frequency f and speed of sound in PPmelt cPP. For the flowing PP melt can be calculated to1f = 2.7 Hz. Since the value is very small the Doppler ef-fect and associated artifacts are considered negligible.

Nevertheless, homogeneity in the flow direction is essen-tial to avoid motion artifacts in the reconstruction. As ex-pected, measurements on PP melts showed more signal loss(impedance mismatch, attenuation) as well as more noisethan in preliminary tests with water. As can be seen in Fig. 5,a higher maximum transmit voltage and a higher gain set-ting is necessary to achieve comparable signal amplitudes inPP to those in water. In this case, the pulse detection is dis-tinctly more challenging. It is possible to penetrate 60 mm offilled PP melt, and a transmission pulse can still be detectedclearly. But its length cannot be determined easily anymorebecause of noise and clutter. Accordingly, the first back wallecho is no longer detectable. Figure 8 shows an example ofan A-scan of two opposing transducers measuring a PP meltfilled with TiO2 and carbon black.

Moreover, the fact that only signals up to an opening an-gle of 60◦ could be evaluated leads to lower quality in theouter region of the images. Figure 9 shows samples of differ-ent extruded rods with a radial filler gradient. As an example,Fig. 10 shows the experimental reconstruction of the sound

Figure 9. Graded filled rods (PP+TiO2+ carbon black), whichhad been examined using the USPT system in the melt.

velocity of pure PP melt as well as of filled melt with incor-porated filler gradation. It can be clearly seen that the outerarea is strongly affected by artifacts, especially after the in-corporation of fillers. The influence of the fillers on the soundvelocity also becomes apparent.

Unfortunately, the gradient cannot (yet) be determined asaccurately as it actually can be created during the manufac-turing process. Furthermore, an inhomogeneous temperaturedistribution cannot be excluded, even though the examinationof pure PP melt exhibits a certain degree of homogeneity. Thereconstruction of the attenuation also proves pretty difficultin the case of PP because the pulse trains cannot be definedclearly and scattering effects have a strong influence as well(see Fig. 8). Moreover, diffraction effects may occur whenfiller agglomerates are present in the direct path between aTRP.

4 Discussion

We have shown that the examination of the local filler dis-tribution is possible with the USPT system. The geometryof the examined graded rods required a tradeoff between thenumber of transducers, transducer size, opening angle as wellas achievable transmission power. As a result, the USPT sys-tem used here merely provides adequate image quality of themiddle part in the reconstruction due to the limitation of theopening angle of maximum 60◦. Efforts to perform real-timecontrol require additional compromises between speed andreconstruction quality. The accuracy of propagation time de-tection using the current algorithm is 0.1–1 µs depending onthe material system and the filler content. Limiting the sig-nals to defined time frames yields time gains. This can be ob-tained by optimizing the sampling frequency and the numberof sample points. However, this is at the expense of time res-olution and the approximate signal propagation times need tobe known. The spatial resolution is expected to be worse than2 mm, when assuming that a perfectly reconstructed point inthe center of the USPT system can only be as small as thetransducers. But this has to be verified in future work.

Filled PP melts could be penetrated with only one UStransducer despite the 60 mm melt channel. However, due toimpedance mismatch and increased signal attenuation, en-

J. Sens. Sens. Syst., 6, 9–18, 2017 www.j-sens-sens-syst.net/6/9/2017/

N. Halmen et al.: Ultrasound tomography for inline monitoring of plastic melts 17

Figure 10. Reconstruction of the experimentally determined soundvelocity of pure (top) as well as filled PP melts with radial fillergradient. Due to the opening angle of 60◦, only the inner area of theimage can be reconstructed satisfactory (dashed marking).

hanced signal amplification is required, which, however, alsoamplifies the noise. Therefore, a larger number of measure-ment repetitions are necessary to receive an adequate SNR.

The artifacts (rings, shadows) caused by the limited open-ing angles can only be removed partially through filtrationand interpolation. Alternatively, an iterative reconstructionprocedure could be used, which, however, requires morecomputing power. Advanced computing methods like algo-rithm parallelization and the use of multicore processors orgraphic processing units can overcome this obstacle and, fur-thermore, even speed up the entire measurement.

The use of matched filter detection was discarded due tothe 1600 TRP, which would have to be treated individuallybecause of different manufacturing and different angles to

each other. For the further development it surely offers thepossibility to enhance the pulse detection. The creation ofvalues between the receivers by linear interpolation provedto be a worthy tool, but it nevertheless needs to be verifiedin a specific setup with a dummy model. The evaluation ofa reconstruction uncertainty is rather difficult at the moment.This is also a topic for upcoming experiments. Wave charac-teristics, diffraction and scattering will be taken into accountfor the next generation of the algorithm. Also an imaging forattenuation is planned. It also has to be investigated whetherthe filler concentration can be locally resolved. Finally, a ref-erence measurement method needs to be developed whichhelps to validate the USPT system by being able to measureat the same system (or dummies) with a well-defined concen-tration distribution.

5 Summary

In this paper we have shown that it is principally possible tolocally detect the spatially resolved distribution of fillers inplastic melts using an USPT system. Considering all influ-ences either by factorization or corrections within the recon-struction algorithm, it is possible to depict the filler distribu-tion in addition to the local temperature distribution. Com-posing the individual sectional images a full 3-D image ofthe rod can be generated subsequently. Thus, US tomographyhas great potential to open up further applications for inlineprocess control of plastic melts. But there is still a well-filledagenda of things to do in further research work.

Information about the Supplement

The supplement contains the original image files used for thesimulation as well as the raw data (Matlab files) of the mea-surements on water, PP melt and filled PP melt, which wereused for this work.

The Supplement related to this article is available onlineat doi:10.5194/jsss-6-9-2017-supplement.

Acknowledgements. The research project VIII/3-3852/45/5 wasfunded by the Bavarian Ministry of Economic Affairs and Media,Energy and Technology. We would like to thank them for thefinancial support.

Edited by: J. CzarskeReviewed by: four anonymous referees

References

Alig, I., Lellinger, D., Lamour, R., and Ramthun, J.: Inline Mon-itoring in the Extrusion Process Using Ultrasonics: Continuous

www.j-sens-sens-syst.net/6/9/2017/ J. Sens. Sens. Syst., 6, 9–18, 2017

18 N. Halmen et al.: Ultrasound tomography for inline monitoring of plastic melts

Monitoring of Quality Parameters, Kunststoffe international, 5,31–33, 2000.

Alig, I., Fischer, D., Lellinger, D., and Steinhoff, B.: Combinationof NIR, Raman, Ultrasonic and Dielectric Spectroscopy for In-Line Monitoring of the Extrusion Process, Macromol. Sy., 230,51–58, doi:10.1002/masy.200551141, 2005.

Alig, I., Lellinger, D., Steinhoff, B., and Fischer, D.: Filler Dis-persion: Ultrasonic Spectroscopy for In-Line Monitoring, Kun-ststoffe international, 5, 60–65, 2006.

Deutsch, V., Platte, M., and Vogt, M.: Ultraschallprüfung: Grundla-gen und industrielle Anwendungen (ultrasonic testing: principalsand industrial applications), Springer publishing house, Berlin,Germany, 1997.

Dinger, F.: Ultraschallverfahren für Prozesskontrolle und Qualität-süberwachung (Ultrasonic method for process control and qual-ity monitoring), Plastverarbeiter, 11, 92–94, 2007.

Haach, V. G. and Ramirez, F. C.: Qualitative assessment of con-crete by ultrasound tomography, Constr. Build. Mater., 119, 61–70, doi:10.1016/j.conbuildmat.2016.05.056, 2016.

Hirt, C., Lang, M., Kretschmer, K., Heidemeyer, P., and Bastian,M.: Grashalm als Vorbild – Gradientenwerkstoffe mit gezieltemVerlauf der Eigenschaftskennwerte (blade of grass as a model– gradient materials with targeted characteristics trend), Kunst-stoffe, 3, 109–112, 2015.

Hochrein, T. and Alig, I.: Prozessmesstechnik in der Kunststof-faufbereitung (process measurement techniques for plastics com-pounding), Vogel publishing house, Würzburg, Germany, 2011.

Hopmann, C., Spekowius, M., Wipperfürth, J., and Schöngart,M.: A concept for non-invasive temperature measurement dur-ing injection moulding processes, AIP Conf. Proc., 1, 1713,doi:10.1063/1.4942274, 2016.

Kak, A. and Slaney, M.: Principles of Computerized Tomo-graphic Imaging, Society of Industrial and Applied Mathematics,doi:10.1137/1.9780898719277, 2001.

Krumova, M., Klingshirn, C., Haupert, F., and Friedrich, K.: Mi-crohardness studies on functionally graded polymer compos-ites, Compos. Sci. Technol., 61, 557–563, doi:10.1016/S0266-3538(00)00228-1, 2001.

Munro, R. G.: Elastic Moduli Data for Polycrystalline Ceramics,NISTIR, 6853, National Institute of Standards and Technology,Gaithersburg, Maryland, USA, 2002.

Praher, B., Straka, K., Usanovic, J., and Steinbichler, G.: Ultrasoundbased monitoring of the injection moulding process – Methods,applications and limitations, AIP Conf. Proc., 1593, 1, 159–162,doi:10.1063/1.4873755, 2014.

Putz, V., Burzic, I., Zagar, B., and Miethlinger, J.: Ultraschall-Messung von Strömungsprofilen in opaker Polymerschmelzewährend der Extrusion (Ultrasound measurement of velocity pro-files in opaque polymer melt during extrusion), tm – TechnischesMessen, 81, 371–380, doi:10.1515/teme-2014-1033, 2014.

Rosen, J. and Gothard, L. Q.: Encyclopedia of Physical Science,Facts on File, New York, 2010.

Samsonov, G. V.: Handbook of the physicochemical properties ofthe elements, IFI-Plenum, New York, USA, 1968.

Schober, G., Heidemeyer, P., Kretschmer, K., Bastian, M., andHochrein, T.: Degree of Dispersion Monitoring by UltrasonicTransmission Technique and Excitation of the Transducer’s Har-monics, AIP Conf. Proc., 1593, 24–28, doi:10.1063/1.4873727,2014.

Schwarzenberg, G.: Untersuchung der Abbildungseigenschafteneines 3D-Ultraschall-Computertomographen (investigationof the imaging properties of a 3D ultrasound computertomograph), PhD thesis, Karlsruhe Series on IntelligentSensor-Actuator-Systems, 6, KIT Scientific Publishing,doi:10.5445/KSP/1000011990, 2009.

Sun, C-K., Chen, L-J., and Chen, H-W.: Plastic wave-guide for ter-ahertz wave, United States Patent, US 7409132 B2, 2008.

Treeby, B. E. and Cox, B. T.: k-Wave: MATLAB toolbox for thesimulation and reconstruction of photoacoustic wave-fields, J.Biomed. Opt., 15, 021314, doi:10.1117/1.3360308, 2010.

Wöckel, S., Arndt, H., Steinmann, U., Auge, J., Dietl, K., Schober,G., Kugler, C., and Hochrein, T.: Statistical ultrasonic char-acterization of particulate filler in polymer compounds, 2016IEEE International Ultrasonics Symposium (IUS), Tours, 1–4,doi:10.1109/ULTSYM.2016.7728747, 2016.

Wu, G., Wen, B., and Hou, S.: Preparation and structural study ofpolypropylene-talc gradient materials, Polym. Int., 53, 749–755,doi:10.1002/pi.1360, 2004.

Zhu, Y-B., Ning, N-Y., Sun, Y., Zhang, Q., and Fu, Q.:A new technique for preparing a filled type of polymericgradient material, Macromol. Mater. Eng., 291, 1388–1396,doi:10.1002/mame.200600249, 2006.

J. Sens. Sens. Syst., 6, 9–18, 2017 www.j-sens-sens-syst.net/6/9/2017/